Title :
An efficient defect compensation scheme for multi-layer neural networks on WSI devices
Author :
Yamamori, Kunihito ; Abe, Tom ; Horiguchi, Susumu ; Yoshihara, Ikuo
Author_Institution :
Fac. of Eng., Miyazaki Univ., Japan
fDate :
6/24/1905 12:00:00 AM
Abstract :
Discusses a high speed off-line defect compensation scheme for trained multi-layer neural networks implemented in WSI devices. Since the partial retraining scheme utilizes the redundancy of neural networks, no additional circuits are needed. The performance of the partial retraining scheme is compared with that of a back-propagation algorithm on a face image recognition problem
Keywords :
compensation; face recognition; fault tolerance; generalisation (artificial intelligence); learning (artificial intelligence); multilayer perceptrons; wafer-scale integration; WSI devices; backpropagation algorithm; defect compensation scheme; face image recognition problem; multi-layer neural networks; partial retraining scheme; redundancy; wafer scale integration; Acceleration; Equations; Image recognition; Information science; Large-scale systems; Multi-layer neural network; Neural networks; Neurons; Parallel processing; Redundancy;
Conference_Titel :
Neural Networks, 2002. IJCNN '02. Proceedings of the 2002 International Joint Conference on
Conference_Location :
Honolulu, HI
Print_ISBN :
0-7803-7278-6
DOI :
10.1109/IJCNN.2002.1005622
